On November 10, 2026, a French expert commission called for stricter oversight of artificial intelligence (AI) assistants, including OpenAI’s ChatGPT, Google’s Gemini, Grok and Meta AI. In a report delivered to President Emmanuel Macron, the experts proposed 18 measures to address risks involving children’s safety, emotional dependence, personal data and hidden advertising. The recommendations are not yet law.
A nine-member expert commission brought together by the French Council on AI and Digital Technology (CIANum) delivered its report to President Emmanuel Macron on Tuesday, November 10. The commission conducted 55 hearings and issued 18 recommendations for safer generative AI, covering consumer chatbots, AI companions and other AI-powered products.
The commission examined how AI assistants may influence people’s decisions, emotions and ability to think independently. Its concerns include chatbots that appear too human, agree with users too readily or encourage people to rely on AI instead of their own judgment. The report places particular attention on children and other users who may be more vulnerable to these risks.
According to the French presidency, generative AI services had approximately 33 million unique users in France in early 2026. The figure highlights the rapid growth of these technologies and helps explain why the commission wants stronger oversight as more people rely on chatbots for information, education, work and personal questions.
One major concern involves AI systems that communicate in ways that make them appear human. Chatbots can respond with friendly messages, supportive language and personalized answers. While these features can improve the user experience, experts believe they may also encourage unhealthy emotional connections, especially among younger users.
The panel recommends stricter limits on AI features that encourage emotional dependence or excessive agreement with users. It also wants developers to reduce design choices that make chatbots appear more human than they are. Some companies have already introduced age-specific safeguards, as TechBonafide previously reported in its coverage of ChatGPT for Teens safety and parental controls.
Protecting children is another important part of the proposals. The commission wants AI services available to minors to block dangerous requests and provide stronger safety features by default, including protections around self-harm, eating disorders and other sensitive topics. The recommendations could also influence France’s position on the proposed EU KIDS Act, which addresses child safety across online services, including AI chatbots and companions.
The commission also recommends stronger controls on AI-powered toys, including safety checks before they reach the market and stricter age limits. Similar concerns have appeared in the United States, as TechBonafide reported in its coverage of California’s proposed AI toys ban.
The commission also wants stronger independent oversight of consumer AI systems. It recommends giving approved researchers and regulators access to relevant information about how these tools are used, trained and monetized. Regular assessments after launch, including checks when models change significantly or risks are reported, could help identify harmful behavior before it affects more users.
Another concern is advertising inside AI conversations. As chatbot companies explore new ways to generate revenue, experts worry that commercial recommendations could influence users without their knowledge. The panel wants AI-generated promotional content to be clearly identified and hidden commercial influence to be prohibited.
Privacy protection is another major concern. The commission warns that AI assistants may infer sensitive details from ordinary conversations, including a person’s health, financial difficulties or political views, even when those details are not directly stated. The experts want users to understand and have greater control over how such information is handled, particularly for commercial purposes. A separate TechBonafide report on sensitive government documents uploaded to ChatGPT illustrates another reason users should be careful about sharing confidential information with AI tools.
The recommendations also propose safety assessments before consumer AI systems become available to the public. Instead of waiting for problems to appear, AI developers could face earlier reviews of potential risks. Additional monitoring would help identify problems after products are released.
The commission also recommends teaching children about AI from primary school and making age-appropriate AI tools available in classrooms. The goal is to help students understand the benefits and limits of AI while encouraging independent thinking instead of automatically accepting chatbot answers.
The commission wants France to improve its ability to detect AI-generated misinformation before it spreads widely. It recommends a rapid response system involving public authorities, researchers and AI companies. The experts also propose a code of conduct for political parties and candidates ahead of France’s 2027 presidential election to discourage the use of AI for misleading or manipulating voters.
However, the proposed measures do not mean ChatGPT, Gemini or other AI assistants are being banned in France. The expert panel recognizes that AI offers important benefits for education, research and productivity. Its goal is to reduce risks while allowing people and businesses to continue using the technology.
The 18 recommendations are not new laws and do not immediately change access to ChatGPT, Gemini or other AI assistants. Some proposals would require action at the European level, where the existing EU AI Act already sets requirements for certain AI systems and providers. The French commission is proposing additional ways to address the everyday risks associated with conversational AI.
For companies such as OpenAI, Google and Meta, the proposed measures could eventually mean broader safety testing, clearer user protections and greater responsibility for chatbot-related harm. The European Union has already introduced separate digital-service obligations for ChatGPT’s search function, as TechBonafide explained in its coverage of ChatGPT’s stricter EU digital-platform oversight.
France’s latest recommendations highlight a growing international debate over how AI assistants should be managed. As these tools become more powerful and widely used, governments face increasing pressure to ensure that innovation does not come at the cost of user safety, privacy and public trust.
